Journal of Convex Analysis 32 (2025), No. 1, 145--168
Copyright Heldermann Verlag 2025
Hybrid Maximum Principle for Regional Optimal Control Problems with Nonsmooth Interfaces
Térence Bayen
Avignon Université, Laboratoire de Mathématiques, France
terence.bayen@univ-avignon.fr
Anas Bouali
MISTEA, Université de Montpellier, INRAE, Institut Agro, Montpellier, France
anas.bouali@inrae.fr
Florent Nacry
Laboratoire de Modélisation Pluridisciplinaire et Simulations, Université de Perpignan, France
florent.nacry@univ-perp.fr
We consider a general Mayer optimal control problem whose controlled dynamics is defined regionally, and, additionally, we suppose that the interface between two regions is nonsmooth in the sense that it is described by a locally Lipschitz continuous function. Our objective is to derive a hybrid maximum principle in this setting. Doing so, we consider a sequence of mollifiers which allows us to approximate uniformly the interface between two regions by a sequence of smooth functions. This makes possible to apply the hybrid maximum principle on a sequence of approximated optimal control problems involving the smooth interface in place of the nonsmooth one. By passing to the limit as the approximation parameter tends to zero, we obtain the desired necessary optimality conditions in the form of a nonsmooth hybrid maximum principle. Our approach relies on the derivation of subgradients of a locally Lipschitz continuous function via mollifiers.
Keywords: Optimal control, Pontryagin maximum principle, hybrid maximum principle, regularization.
MSC: 34A38, 49A15, 49J53.
[ Fulltext-pdf (192 KB)] for subscribers only.